To compare arm-ergometry and treadmill supervised exercise training on cardiorespiratory fitness and walking distances in patients with peripheral artery disease (PAD). ARMEX was a single-center, single-blinded, parallel group, non-inferiority trial enrolling symptomatic patients with PAD. Patients were randomized (1:1 ratio) to a 12-wk arm-ergometry (AEx) or standard treadmill (TEx) supervised exercise training protocol. The powered primary end point was the change in peak oxygen uptake (VO2) at 12 wk, measured on a treadmill cardiopulmonary exercise test (CPX). Secondary outcomes included changes in VO2 at the first ventilatory threshold (VT-1), ventilatory efficiency (ratio of minute ventilation [VE] to carbon dioxide production [VCO2], VE/VCO2), walking distances by CPX and 6-min walking test (6MWT), and self-reported walking limitations. Fifty-six patients (66±8yr; 88% male) were randomized (AEx, n =28; TEx, n =28). At 12 wk, VO2peak change was not significantly different between groups (0.75 mL/kg/min; 95% CI, -0.94 to 2.44; P=.378), despite a significant increase only in AEx. VO2 at VT-1 improved in both groups without between-group differences, and VE/VCO2 slope improved more in AEx. The TEx attained greater improvements in walking distance by CPX (121.08 m; 95% CI, 24.49-217.66; P=.015) and 6MWT (25.08 m; 95% CI, 5.87-44.29; P=.012) and self-perceived walking distance. Arm-ergometry was noninferior to standard treadmill training for VO2peak, and treadmill training was associated with greater improvements in walking distance. Our data support the use of treadmill as a first-line choice in patients with PAD to enhance walking capacity, but arm-ergometry could be an option in selected patients.